Speaker
Description
Surveys are facing pressures to shorten questionnaires: Long questionnaires are associated with low response rates, poor response quality, and are particularly considered inappropriate for the increasingly popular online mode. This is why survey designs with planned missing data, such as split questionnaire designs, are becoming more and more common in large-scale social surveys: They help reducing survey length by administering varying components of the whole questionnaire to each respondent. However, imputation may be needed to obtain reasonably analyzable data with such a design. Yet, these data can be difficult to impute due to common features of social survey data, such as low correlations, predominantly categorical data, and relatively small sample sizes available to support imputation models with many potential predictor variables.
In this presentation, I will discuss findings from a series of Monte Carlo simulation studies in which split questionnaire designs are simulated using real social survey data from the German Internet Panel and subsequently imputed. Estimates based on the imputed data are compared to population benchmarks to determine their accuracy. In the course of these studies, several different strategies regarding the design of the planned missing data and their imputation are examined in their effects on the accuracy of estimates.
